High-resolution remote sensing images such as GF-1/GF-6 PMS with a spatial resolution of 2 meters after fusion and ZY3-02 satellite are used to carry out research on the refined classification model of rice. Four classification methods of Maximum Likeli-hood, Mahalanobis Distance, Neural Network and Support Vector Machine (SVM) were used to identify single-cropping late rice and double-cropping late rice base on three different data combinations of single-temporal image, multi-temporal images, and multi-tem-poral combined with texture feature.Through the experimental research, a relatively good combination model of rice identification was obtained, which has certain guiding significance for the subsequent identification and extraction of rice information based on domestic high-resolution satellite images.
维普
万方数据